Administration of an aldose reductase inhibitor induces a decrease of collagen fluorescence in diabetic rats

Administration of an aldose reductase inhibitor

induces a decrease of collagen fluorescence in

diabetic rats.

G Suárez, … , A L Oronsky, J A Goidl

J Clin Invest.

1988;82(2):624-627. https://doi.org/10.1172/JCI113641.

As a consequence of an increased flux through the sorbitol pathway fructose levels rise in

various tissues in diabetes. Also, in vitro nonenzymatic fructosylation of protein induces the

generation of fluorescence at a rate 10 times greater than glucosylation. The administration

of sorbinil, an aldose reductase inhibitor known to lower tissue fructose concentration, to

experimental diabetic rats led to a decrease in the fluorescence related to advanced

Maillard products in their skin collagen. This effect is consistent with the in vivo occurrence

of nonenzymatic fructosylation of collagen. A potential pathogenetic role for this

posttranslational modification in diabetic complications should be considered.

Research Article

Find the latest version:

Administration

of

an

Aldose Reductase Inhibitor

Induces

a

Decrease

of

Collagen Fluorescence

In

Diabetic Rats

GerardoSuarez, Rama Rajaram, Kailash C. Bhuyan,* Arnold L.Oronsky,and JoAleneGoidl*

Milton and MiriamPetrieArthritis ResearchLaboratory, Department ofOrthopaedics, and *Department ofOphthalmology, MountSinai School ofMedicine, NewYork, New York 10029; tMedical ResearchDivision,Lederle Laboratories,

DivisionofAmericanCyanamid Company, PearlRiver, NewYork10965

Abstract

As

a consequence

of

an

increased

flux

through the sorbitol

pathway fructose levels rise in

various tissues in diabetes.

Also, in

vitro

nonenzymatic

fructosylation of protein induces

the

generation of

fluorescence

at a rate

10 times

greater

than

glucosylation. The

administration of

sorbinil,

an

aldose

reduc-tase

inhibitor known

to

lower

tissue

fructose concentration,

to

experimental diabetic

rats

led

to a

decrease in the

fluorescence

related

to

advanced

Maillard products in their skin collagen.

This

effect is consistent with the in vivo

occurrence

of

nonen-zymatic

fructosylation of collagen.

A

potential pathogenetic

role

for this

posttranslational modification

in

diabetic

compli-cations should be

considered.

Introduction

Despite

significant improvements

in the control

of

blood

glu-cose

levels

by

diet

or

drug therapy

the

long-term

complications

of

diabetes

are

still

leading

causes

of

death and

morbidity (1,

2).

Among

the various biochemical

processes

that have been

proposed

as

underlying

the

pathogenesis

of these

complica-tions,

nonenzymatic

glucosylation'

of proteins

has been the

focus

of

increasing attention

during

the last

decade

(3-5).

The

term

nonenzymatic

glycosylation

collectively

denotes

a

mul-tireaction

pathway, also termed the

Maillard reaction,

whereby

reducing

sugars react

with

protein amino

groups

and

form

Schiff base-mediated

adducts that

rearrange

into

more

stable

glycoconjugates.

If the

sugar

is

glucose

the

rearrange-ment

is known

as

the

Amadori

rearrangement

and the

glyco-conjugate is

a

ketoamine-linked I-deoxy

hexose. The

products

of

the rearrangement

undergo

further

transformations

which

lead

to

the

emergence

of poorly characterized fluorescent

structures.

Recent

evidence (6)

supports

the

view that these

Apreliminaryaccountof these results has beenpresentedin

Kashiko-jima,Japan(Suarez, G., R.Rajaram,K.Bhuyan,A.L.Oronsky,and J.A.Goidl. 1986.International SymposiumonPolyolPathwayandits Role inDiabeticComplications. 38).

Address all correspondenceto Dr.GerardoSuarez, Department of

Biochemistry,NewYorkMedical College, Valhalla, NY 10595.

Receivedfor publication8 June 1987 andinrevisedform22 Febru-ary 1988.

1.Throughout thetextthe word"glycosylation"has beenemployedto

denotethenonenzymaticproteinmodification byany sugarwhereas "glucosylation" denotesexclusivelythereaction of glucose with pro-tein.

fluorescent moieties mediate the establishment of crosslinks

between

protein molecules which

may

underlie

some

pathoge-netic

events

in the

complications

of

diabetes

(7).

Alternative hypotheses

to

explain the complications

of

diabetes

are

based

on

the

operation of the sorbitol pathway.

This is

a

metabolic

shunt, the

net

result

of which is

the

con-version of glucose

to

fructose (Scheme I) with

the

formation

of

sorbitol

as an

intermediate

metabolite. Reduction of

glucose

is

catalyzed by the

NADPH-requiring

enzyme

aldose reductase

(alditol:

NADP

oxidoreductase,

E.C. 1.1.1.21). The resulting

sorbitol

can

be oxidized

enzymatically

to

fructose through the

action of sorbitol dehydrogenase

(L-iditol:NAD

oxidoreduc-tase,

E.C.

1.1.1.14), with

NAD+

serving

as a coenzyme. An

explanation of the long-term

complications of

diabetes such

as cataracts

and

peripheral

neuropathy (8, 9) is based

on

the

accumulation of sorbitol

within

cells.

Sorbitol accumulation

occurs

because the cell membrane

is impermeable

to

sorbitol,

and

thus,

a

higher

flux

through

the

sorbitol

pathway,

as

in

diabetes,

would result in

a

rise in the

intracellular

level

of the

polyol.

This,

in

turn,

would

induce

an

increase

in the osmotic

pressure

leading

to

the

transfer of

water to

the

intracellular

compartment,

ionic

leakage,

and cell damage.

In contrast to

the role

of sorbitol

the second

step

in the

sorbitol pathway, the

generation

of

fructose,

has

received,

to our

knowledge,

almost

no

attention although fructose levels

have been shown

to

increase

up to

23-fold

in

tissues of diabetic

animals

where the

sorbitol

pathway is active (10).

Numerous

other studies have documented elevated

intracellular

concen-trations

of fructose

in

diabetic animals

(1

1-13).

The

concept

that the

sorbitol

pathway

might mediate diabetic

complica-tions

by

mechanisms other than those related

to

the "osmotic

hypothesis" described

above

was

suggested

to us as a

result

of

considerations of the

following experimental findings:

incuba-tion

of BSA with 0.5

M

fructose led

to

the

generation

of

pro-tein-bound fluorescent

products

at arate

10 times

greater

than

when

glucose

was

used

as

the

glycosylating

sugar

under the

same

conditions

(14). We,

therefore,

concluded that fructose

was a

much

more

efficient glycosylating

sugar

than

glucose.

Based on

this

we

hypothesized

that

nonenzymatic

fructosyla-tion

might

occur

in vivo

and,

if

so,

might

mediate

complica-tions of diabetes

that

depend

on

the

operation

of the sorbitol

pathway.

To test

this

hypothesis

a

study

was

undertaken that

com-pared

protein-bound

fluorescence in skin from diabetic

rats

treated

with the aldose

reductase

inhibitor,

sorbinil

([+]6-fluoro-spiro-[chroman-4,4'-imidazolidine]-2',5'-dione),

with

that

found in the skin of untreated diabetic animals.

Itwas

conjectured that

protein-bound

fluorescence

resulting

from

nonenzymatic

fructosylation

would be

affected by sorbinil

but notthat

which

results

from

glucosylation,

since this

treatment would have no

effect

onthe

glucose levels

(12,

15,

16).

J.Clin.Invest.

©TheAmericanSocietyfor ClinicalInvestigation,Inc.

0021-9738/88/08/0624/04

$2.00

NADP

NADPH+ D-GLUCOSE

SORBITOL NA

orbitol NAhydrogenHse

D-FRUCTOSE NANDH+H' Scheme 1.

Methods

Two groups of maleSprague-Dawleyrats

weighing

anaverage of 160 g wererendereddiabeticbyasinglesubcutaneousinjectionof

strepto-zotocin (80 mg/kg body

weight).

Oneof these groupsof diabeticrats

wasfed adiet supplementedwith sorbinil(400 mg/kgof rodent labora-torychow), whichwaskindlyprovided by

Pfizer,

Inc.,

Groton,

CT. After 56 d oftreatmentthe rats were killedby cervical dislocation

immediately after withdrawal of blood from the retrobulbar blood vessels.

Solubilization ofskin

collagen by

bacterial

collagenase.

The skin of eachanimalwasremovedby careful dissectionand stored frozenat -80°C until used. Portions ofskinwere cut into small

pieces

and washedtwice with 0.05 Mphosphate, pH7.4,

containing

0.1 M NaCl (PBS),0.01 MN-ethylmaleimideand0.01 M EDTAbyslow

overnight

stirringandsubsequentdecantation. After removal of hair the residue wassuspended in 10 mlPBS andhomogenizedwithaPolytron ho-mogenizerat4°C.Thehomogenatewascentrifugedat10,000g for 10 minat4°C and the supernatant discarded. Thepelletwaswashedwith 30 ml of coldwaterthreetimes.Toeach washedpellet200 mlofa1:1,

vol:vol,ethanol/ether mixturewasadded andstirred

overnight.

The

ethanol/etherextract wasdiscarded andthe residuewaswashed with water by centrifugation at 12,000 g. The washed pellet was

resus-pended in 5 ml of0.05 Mborate, pH7.4,

containing

I MNaCIfor

preequilibration,andresuspendedin 2 ml ofamixturecontaining0.05 Mborate, pH7.4,50mM

CaCl2,

20 mM

N-ethylmaleimide

and 155 U

of Clostridium

histolyticum collagenase,

form III (Advance

Biofac-tures,Lynbrook, NY).

Digestion

wasachievedby

stirring slowly

over-nightat37°C. Theincubationmixturewas

centrifuged

at12,000g for 30 min and assayswereconducted in supernatant

aliquots.

Fluores-cence measurements werecarriedoutina

recording

spectrofluorome-ter(204 A; Perkin-Elmer

Corp.,

Pomona,

CA)

using

an excitation

wavelengthof350 nm,anemissionwavelengthof 410 nm, and slit widthof 10nmfor both excitation and emission. The instrumentwas

adjustedsothat theemissionofthe standardNo.4from the manufac-turers (tetraphenylbutadiene)was70at a

_{photomultiplier}

_gain

of 3 anda

sensitivity

of 0.3. Fluorescence valueswere normalized with respecttothehydroxyprolinecontentof the

collagenase

digest

as de-terminedaccordingtoWoessner

(17).

Fluorescencemeasurementsof

all thesampleswereconductedon a

_{single day.}

Incubation

ofBSA

with sugars. BSA(fatty acid-free, Sigma Chemi-cal Co., St. Louis, MO), used as a model protein, was incubated at a concentration of 6 mg/ml in 0.05 M sodium phosphate 0.1 M NaCi, pH 7(PBS), containing either 0.5 M glucose or 0.5 M fructose at 37°C for 32 d under sterile conditions in a humidified atmosphere. Subse-quently, theincubationmixtures were dialyzed vs. 400 vol of PBS for 4 d with several changes of the effusate. Then, the dialyzed protein solu-tions were centrifuged at 10,000 gat4°C for 20min,the supernates were passed throughultrafiltration membranes (Millipore/Continental WaterSystems,Bedford,MA) and storedat4°C. The sterile solutions wereprecipitated with 5 vol of coldtetrahydrofuranand the precipi-tates werecollected by centrifugationat3,500 gfor 10 min and washed threetimes withtetrahydrofuran. The washedprecipitateswere sus-pended in PBS anddialyzedvs.100volof PBS with fourchangesofthe effusate.

Fluorescenceemission spectra. Collagenasedigestsobtained as

de-scribedin thelegendtoFig. 1werescanned in thePerkin-Elmer spec-trofluorometer. Full pen deflection correspondedto 1 V. Excitation wavelength was 350 nm. All samples were taken to a concentration of hydroxyproline of 400

jg/ml.

Nonenzymatically glycosylated BSA solutions, obtained after in-cubation with sugarsasdescribedabove,weretakento aprotein con-centration of 0.365 mg/ml (as determined by the Lowry method) and theemission spectra taken as described.

Results

Relevant blood glucose values

and

body

weight

changes

of

the rats are

summarized in Table

I.

Control

rats

gained a mean of

150.8

g

during the period of observation while diabetic rats

lost

amean

of 16.18

g

(untreated)

and

22.19

g

(sorbinil-treated).

Fig.

1

shows

the results

on the

fluorescence associated

with the material

that

was

solubilized from the skin by purified

bacterial

collagenase. Preliminary

surveys on

samples

from 50

rats

revealed that the solubilized material had an amino

acid

composition almost identical

to

that of

type I collagen ob-tained from rat skin

according to the method of Chandrakasan

et al.

(18).

A

minor proportion of cysteine indicated

that a small percentage

of

type III

collagen

(19), also present in the

skin,

was

likewise solubilized. In the same preliminary study

no

significant differences were found in the yield of

collagen-ase-released

hydroxyproline-containing

material between

dia-betic and control animals.

Control

ratskin

yielded 94.2±22.44

SD nmol of

hydroxyproline/mg wet tissue, while diabetic rat

skin

yielded 90.97±21.46 SD nmol (control

vs.

diabetic

P

<

0.6). The normalized fluorescence associated with skin

col-lagen

of

the untreated diabetic rats (1 17±7) was almost twice

that of control

rats

(63±5). After

treatment

with sorbinil

fluo-rescence was

significantly (P < 0.003) reduced (87±4).

Table

L Dataon

Experimental Animals

Experimentalgroup Plasmaglucose(mean±SE) Body weight (mean±SE)

g

Days afterstreptozotocin

injection

3 56 0 56 Control 135.6±2.3

mg/dl

ND 162.2±1.3 313±11.03

n= 12 n= 12 n= 12

Diabetic,untreated 513.9±6.4mg/dl 515.8±15.4

mg/dl

159.6±1.1 143.4±6.2

n= 14 n= 14 n= 14 n= 14

Diabetic,sorbinil-treated 534.1±9.4

mg/dl*

494.1±28.2mg/dlt 157.6±1.3 129.5±3.6§

n= 14 n= 14 n= 14 n= 12

nmay differ from number of animalsatthestartof the

experiment

duetodeathor

inadequate

sample. *Not

significantly

different from

un-treateddiabeticrats at P. 0.084. $ Not

significantly

different fromuntreateddiabeticratsP.0.5. §Not

significantly

different from

un-treateddiabeticrats at P. 0.077.

U Figure 1. Normalized relative

140 fluorescence of

collagenase-sol-12 ubilized material of normal

0 20 . . _ _{and diabetic rats. Fluorescence}

LL

loo _ _ emission

(excitation

wave

W 100

> _{; length,350 nm) was measured}

, 80

_

in

samples

of solubilized skin

o

._

.obtainedasdescribedunder

N 60 _ . . _ Methods. Meanvalues were

4Q

63.7, 87.6,

and 117.2for the 0

control,

sorbinil-treated dia-20 betic and diabetic group of

0T

WI

rats, respectively. P < 0.003

CONTROL DIABETICS DIABETICS for thediabeticvs.

sorbinil-TREATED WITH

SORBINIL treateddiabetic group.

Fluorescence spectra

were

analyzed

to

characterize the

fluorophors associated

with

collagen.

For

this purpose

samples

were

chosen

with

a

fluorescence emission closest

to

the

mean

value

of

the

respective experimental

group.

Fig.

2

illustrates

the

emission

spectra

of the collagen

solubilized from the skin

of normal,

diabetic,

and

sorbinil-treated diabetic

rats.

Sorbinil

treatment led

to a

diminished

emission

maximum,

compared

with untreated

diabetic

rats,

while

the

qualitative

features

of

the spectra

remained

unaltered.

Furthermore,

the

emission

spectra

of

the

collagenase-solubilized

material

was

almost

in-distinguishable from

the

one

obtained in the BSA

samples

that

had been

incubated with fructose

at

370C for 32 d.

Fig.

2 also illustrates the fact that

although

the fluorescence

of

nonenzymatically fructosylated

BSA

was

qualitatively

un-distinguishable

from that of the

nonenzymatically

glucosyl-ated

protein,

as

is

the

case

for collagen,

its relative

intensity

was much

higher (about

10-fold).

Discussion

From these data

it

was

obvious

that sorbinil

treatment

induced

a

highly significant reduction in collagen-bound fluorescence

A B Figure 2. (A)

Fluores-o0 _{cenceemissionspectra}

9 ofratskin

samples

solu-bilizedby collagenase.

U 8

Samples

wereobtained

z

W

l l asdescribed in the

leg-,^J 7_ X \ endto

Fig.

1.a,

control;

l _{b, sorbinil-treated}

dia--J 6 betic; c, diabetic. With 2 5- theinstrumentsettings

corresponding to the

gr4- _{I \} I I measurementsreferred

N toinFig. 1,samplesa,

< 3

bl

\\ \ b,andcshowed

nor-or2 X~k \

\malized

emission values

z

a

of60,

88,

and 1

13,

re-b

_spectively.

_(B)

Fluores-a cenceemission spectra

0

_...0

400 450 500 550 35040045450500~~55060Wenyalalofo

nonenzymatically

WAVELENGTH(nm)

glycosylated

BSA.BSA wasincubated for 32 d at

370C

without (a) sugar,(b)orin the presence of 0.5Mglucose, (c) orfructose.Protein concentration in each samplewas0.365mg/ml. (Seetextfor details of thepreparations.)

related to the later stages of the Maillard reaction. The dose of

sorbinil used in this study has

previously

been

shown to

de-crease

fructose

levels in

tissues

where the

sorbitol pathway is

active

(12, 15, 16).

Since

the

inhibitor did

not

affect the blood

glucose levels

of the diabetic rats (Table

I),

as previously

re-ported

(12,

15,

16), the decreased collagen

fluorescence

cannot

be

explained by

a

reduction in

the extent of nonenzymatic

glucosylation. The results are, however, consistent with a

de-creased

generation of fluorophors

that

arise from

nonenzyma-tic

fructosylation

of

proteins.

It

should be

noted, however, that

administration of aldose

reductase

inhibitors

has been shown

to

decrease

sorbitol

values more

efficiently

than

fructose

levels

(12,

15,

16).

This may explain, in part, the inability of sorbinil

to reverse

collagen-associated fluorescence

completely.

An-other

explanation for

the

incomplete

reversal

of

the

high

col-lagen

fluorescence after sorbinil

treatment is the probable

coexistence

of

nonenzymatic glucosylation

and

fructosylation,

which would be expected to occur under conditions of

hyper-glycemia

uncorrected by sorbinil treatment. As already

men-tioned,

nonenzymatic glucosylation

also leads to the

genera-tion

of

protein-bound fluorescent moieties.

The extent

of

nonenzymatic glycosylation, which is a very

slow

reaction, for

a

given

protein

is

largely

dependent on

its

half-life.

Thus, it could be

hypothesized

that the decreased

fluorescence

induced

by

sorbinil

is

a

reflection of

increased

collagen turnover.

This is highly improbable,

based on the

almost

identical growth

curves

of

the

two

groups

of diabetic

rats

and

on

the

documented lack

of effect of sorbinil

on

colla-gen

synthesis (20).

These

considerations

have led us to the view that our

re-sults

provide

presumptivc

evidence

for

the occurrence

of

non-enzymatic fructosylation

in

vivo.

It

has been

proposed

that

aldose reductase

inhibitors

might

have

beneficial effects

through

a

mechanism

that

involves

a

replenishment of

re-duced

glutathione, which

would lead

to

decreased

disulfide

bridge-mediated

protein

crosslinks

upon

nonenzymatic

glyco-sylation (21). Fluorophors

resulting

from

nonenzymatic

glu-cosylation, however,

appear

to

be

generated

at a

higher

rate

under

nonoxidative conditions

(22),

soour

results

cannot

be

satisfactorily

explained by

this

hypothesis

if

fructosylation

proceeds by

analogous mechanisms.

The

sorbitol pathway is generally thought

to

be

an

intra-cellular

pathway,

while the present

study

deals

with the

struc-tural

modification of collagen, which is

an

extracellular

pro-tein

in

its

mature

form.

However,

a recent

report

proposed

extracellular

sorbitol accumulation

as a

pathogenetic

determi-nant in the

syndrome of limited

joint mobility (23),

one

of

the

sequelae of

long-standing juvenile

diabetes (24).

Increased

ex-tracellular

polyol concentration

would lead

to a

higher degree

of collagen

hydration,

which

might

underlie

the

pathophysiol-ogy

of virtually

all

the

complications

of

diabetes in

collagen-containing

organs. The

general

validity

of the

"collagen

hy-dration"

hypothesis (25, 26)

must

be

questioned

since

nerve

water content of

peripheral

nerve

has been

found unaltered in

experimental diabetic

rats

(27). Also,

the

nonenzymatic

fruc-tosylation

concept has the

potential

to

provide

an

explanation

of

the

pathology associated

not

only

with

collagenous

proteins

but also

with

noncollagenous

proteins,

such as

lens

crystallins

and

red cell

membrane

proteins,

aswell.

Although

fructose is

generated

intracellularly,

there

is evidence that

significant

dif-fusion of the

sugar

into the culture

medium

occurs

in

organ

sorbinil

was

effective

at

decreasing

the concentration of

sorbi-tol

in human endothelialcells cultured in the

presence

of

high

glucose concentrations.

This

finding

demonstrates that

the

sorbitol pathway is

present in all kinds of vascular tissue

and

that the levels

of

sorbitol,

and

presumably

of

fructose,

can

be

modulated

by sorbinil. Leakage

of fructose fromvascular tis-sue,

where

the sorbitol

pathway

is

active,

tothe extracellular

collagenous matrix

of the skin would

be favored by

the

high

degree of

vascularization of this tissue

(30).

In summary, our

results

are

compatible with

the in

vivo

occurrence of

protein nonenzymatic

fructosylation

and

war-rant

further exploration aimed

at

providing direct

evidence of

this posttranslational modification,

such as identification

of

the corresponding advanced Maillard

reaction

products. If

nonenzymatic

fructosylation

represents a

major event,

then

the value of protein-bound

fluorescence as a clinical marker

(31) and its pathophysiological significance should be

reas-sessed.

Acknowledgments

Supportedin partbyAmerican Heart Association grant

82-943,

and by the Milton and Miriam Petrie Fund and Lederle Laboratories.

References

1. Krolewski, A. S., J. H. Warram, and A. R. Christlieb. 1985.

Onset,course, complications, andprognosisofdiabetes mellitus. In Joslin'sDiabetesMellitus.A.Marble,L.P.Krall, R. F. Bradley,A.R.

Christlieb,and J.S. Soeldner, editors.Lea andFebiger, Philadelphia.

251-277.

2. Entmacher, P., L. P. Krall, and S. N. Kranczer. 1985. Diabetes mortality fromvital statistics.InJoslin's DiabetesMellitus. A. Marble, L. P. Krall, R. F. Bradley,A.R.Christlieb,and J.S. Soeldner, editors. LeaandFebinger, Philadelphia.

3. Cerami, A.,V. J.Stevens,and V. M. Monnier. 1979. Role of

nonenzymatic glycosylation in the development of the sequelae of

diabetesmellitus.Metab. Clin.Exp.28(Suppl. 1):431-437.

4. Kennedy, L., and J. W. Baynes. 1984. Non-enzymatic glycosyla-tion and the chroniccomplicationsofdiabetes:anoverview.

Diabeto-logia.26:93-98.

5.Brownlee,M.,H.Vlassara, andA.Cerami.1984. Nonenzymatic

glycosylation and the pathogenesis ofdiabetic complications. Ann. Intern.Med. 101:527-537.

6.Brownlee, M., H. Vlassara, A.Kooney, P.Ulrich,andA. Cer-ami. 1986. Aminoguanidine prevents diabetes-induced arterial wall

protein cross-linking. Science(Wash.DC).232:1629-1632.

7. Schnider, S. L., and R. R. Kohn. Effects of age and diabetes mellitusonthesolubilityandnonenzymatic glucosylation of human skincollagen.J. Clin.Invest.67:1630-1635.

8.Kinoshita,J.H., S.Fukushi,P.Kador, and L.0.Merola. 1979. Aldose reductase in diabetic complicationsof the eye. Metab. Clin. Exp.28(Suppl. 1):462-469.

9. Gabbay, K.H., L.0.Merola, and R.A. Field. 1966. Sorbitol pathway: presence in nerve and cord with substrate accumulation in diabetes. Science(Wash.DC). 151:209-210.

10. Gabbay, K. H., and J. H. Kinoshita. 1972. Mechanisms of development and prevention of cataracts. Isr. J. Med. Sci.

8:1557-1561.

11. Heath, H., and Y. C. Hamlett. 1976. The sorbitol pathway: effect of streptozotocin induced diabetes and the feeding of a sucrose-richdietonglucose, sorbitol and fructose in the retina, blood and liver of rats.Diabetologia. 12:43-46.

12.Poulsom, R.,R. P.Boot-Handford,and H. Heath. 1983. The effects oflong-termtreatmentof streptozotocin-diabetic rats withan

aldose reductase inhibitor.Exp. EyeRes. 37:507-515.

13.Tomlinson,D.R.,J.Townsend,and P.Fretten. 1985. Preven-tion of defective axonal transport in streptozotocin-diabetic rats by treatmentwith"statil"(ICI 1228436),and aldose reductase inhibitor. Diabetes. 34:970-972.

14.

Sua'rez,

G., J. Novick, and A. L. Oronsky. 1984. Fed. Proc. 43:2022.(Abstr.)

15.Poulsom, P., D. J. Mirrlees, D. C. N. Earl, and H. Heath. 1983. The effects ofanaldose reductase inhibitor upon the sorbitolpathway,

fructose-1-phosphateandlactate in the retina andnerveof streptozo-tocin-diabeticrats.Exp. EyeRes. 36:751-760.

16.Stribling, D.,D. J.Mirrlees,H. E.Harrison,and D.C. N. Earl. 1985. Properties of ICI 128, 436, a novel aldose reductaseinhibitor,

and its effectsondiabeticcomplicationsin therat.Metab. Clin. Exp.

34:336-344.

17. Woessner, J. F., Jr. 1961. The determination of hydroxyproline in tissue and protein samples containing small proportions of this imino acid. Arch. Biochem.Biophys.93:440-447.

18.Chandrakasan, G.,D. A. Torchia, and K. A. Piez. 1976.

Prepa-ration of intact monomeric collagen from rat tail tendon and skin and the structure of the nonhelical ends in solution. J. Biol. Chem. 251:6062-6067.

19. Chung, E., and E. J. Miller. 1974. Collagen polymorphism: characterization of molecules with the chain composition

[1(III)13

in human tissues. Science (Wash. DC). 183: 1200-1201.

20. Li, W., M. Khatami, and J. H. Rockey. 1985. The effects of glucose andanaldosereductase inhibitor on the sorbitol content and collagen synthesis of bovine retinalcapillarypericytesin culture. Exp. Eye Res. 40:439-444.

21. Bitensky, M. W., A. Kowluru, and R. A. Kowluru. 1986. In-creased cross linking of proteins in diabetes. Fed. Proc. 45:1837.

(Abstr.)

22.Ahmed, M. U., S. R. Thorpe, and J. W. Baynes. 1986. Identifi-cation ofN-carboxymethyllysine as a degradation product of fructose-lysine in glycated protein. J. Bio. Chem.261: 4889-4894.

23.Eaton, R. P., W. L. Sibbitt, and A. Harsh. 1985. The effect ofan aldosereductase inhibiting agent on limited jointmobilityin diabetic patients. J. Am. Med. Assoc. 253:1437-1440.

24.Rosenbloom, A. L., J. H. Silverstein, D. C. Lezotte, K. Richard-son, and M. McCallum. 1981. Limitedjoint mobilityin childhood diabetesmellitus indicates increased risk for microvascular disease. N. Engl. J. Med. 305:191-194.

25. Eaton, R. P. 1986. Aldose reductaseinhibitionand thediabetic syndrome of limited jointmobility: implications for altered collagen

hydration.Metab. Clin. Exp.35(Suppl. 1): 119-121.

26. Eaton, R. P. 1986. The collagenhydrationhypothesis: a new

paradigm for thesecondary complicationsofdiabetes mellitus. J.

ChronicDis.39:763-766.

27.Tomlinson, D. R., G. B. Willars,andJ. P. Robinson. 1986. Prevention of defects ofaxonaltransportin experimentaldiabetesby aldosereductaseinhibitors. Drugs. 32(Suppl. 2): 15-18.

28.Morrison,A.D., R.S. Clements, Jr., and A. I. Winegrad. 1972. Effects of elevated glucose concentrations on the metabolism of the aortic wall. J. Clin. Invest. 51:3114-3123.

29. Lorenzi, M., S. Toledo, G. R. Boss, M. J. Lane, and D. F. Montisano. 1987. The polyol pathway and glucose 6-phosphate in human endothelial cells cultured in high glucose concentrations.

Dia-betologia.30:222-227.

30. Ham,A.W.1974.Histology.7th ed. J. B. Lippincott, Philadel-phia. 598.

31. Monnier,V.M., V.Vishwanath, K. E. Frank, C. A. Elmets, P. Dauchot, andR.R. Kohn. 1986. Relation between complications of type Idiabetes mellitusandcollagen-linked fluorescence. N. Engl. J. Med. 3 14:403-408.